Projects

A scalable SaaS platform offering a REST API for real-time extraction and enrichment of Google Maps business data, designed for lead generation and market analysis.

• Extracts over 185 data points from up to 500 unique businesses per search query, providing comprehensive location-based intelligence.
• Features multi-level data enrichment to transform raw search results into qualified leads by adding social profiles, emails, and key decision-maker details.
• Built as a commercial API with a pay-per-success credit system, developer-friendly documentation, and robust infrastructure for reliable, high-volume scraping.

• Search engines such as Google, Bing: Extract search results to CSV from a range of pages or all pages.
• Complete data extraction from social media, real estate, job boards, and mobile applications.
• Amsoil Extractor - Extracts engine details and associated products based on year, model, and make. Uses API to create articles on a WP blog with affiliate codes (LLM integrated).

A sophisticated Telegram bot for on-demand text-to-speech conversion with customizable voice options and robust user management.

• Built with Python, Telegram API, and ElevenLabs API for high-quality TTS conversion.
• Features user registration, access control, usage tracking, and command-based interaction.
• Uses MongoDB to store user data, manage access, and log prompts for analysis.

• Managed 500k+ records in OracleDB, developed web apps with JavaScript, PHP, and SQL.
• Secured server with UFW, optimized performance with Nginx, and used Certbot for SSL.
• Admin and user-level access control, allowing directors to grant user-level access to officers.

• Secured client-server connection with Asterisk PBX on Ubuntu.
• Automated PBX server monitoring and alerts using Bash scripts.
• Developed on Raspberry Pi 4 with optimized network infrastructure.

Research project co-authored with Y A Joarder. Focused on applying OpenStack to build a network security technology platform.

A book recommender system using user-based Collaborative Filtering Algorithm.

• Dataset includes 278,858 users, 1,149,780 ratings, and 271,360 books.
• Content-based filtering provides popular book recommendations to users.

A social platform REST API with a voting system, JWT authentication, and authorization.

• User login, registration, and authentication with JWT tokens.
• PostgreSQL database with secured API endpoints.
• SQLAlchemy and Alembic for CRUD operations based on Pydantic models.
• Unit testing with PyTest, API testing with Postman, and Dockerized deployment.
• CI/CD pipeline with GitHub Actions for automated deployment, optimized with Nginx.

A deep learning model for predicting brain tumors from MRI images using TensorFlow CNN. Achieved 98% accuracy across four classes: meningioma, glioma, pituitary tumor, and no tumor.

• Data scaling and normalization using Keras.
• Images processed with NumPy and OpenCV (150x150 resolution).
• Used EfficientNetB0 with transfer learning from the ImageNet dataset.
• TensorBoard, ModelCheckpoint, and ReduceLROnPlateau for callbacks.

PhD Task: Geometrically transform stereoscopic 2D images based on photographic composition rules.

• OpenCV and NumPy for image processing.
• Hybrid feature matching with ORB and Viola-Jones detectors.
• Implemented the rule of thirds with Haarcascade classifier.

Contact: Y A Joarder

Globally standardized self-taught study plans on GitHub covering Computer Science, Data Science, Machine Learning, Web Development, AI, Higher Education, and Data Structures & Algorithms. Curated resources for self-learners worldwide.